it235 Test Coverage T2 (Coverage Evaluation)
Only after G2 PASS, before T3. Evaluate "is T1 enough", do not execute environment verification.
Integrated from
ac-coverage-t2-squad. Material assembly rules seemultica-test-orchestration"Assemble from Comments".
5 Required Items (assemble from comments; stop if any missing)
| # | Required | How it counts as complete in comments |
|---|---|---|
| 1 | Numbered AC- original text | AC pasted in comments/description, or openable Jira/PRD link |
| 2 | T1 Confluence link | t1-cases Confluence URL (preferred); or already-imported JIRA case set |
| 3 | G2 PASS evidence | Leader writes G2 PASS |
| 4 | Frontend + backend changes | SHA + file list, or MR/PR link |
| 5 | Post-implementation API contract | Apifox/OAS link, or N/A (reuse xxx) |
| Optional | Note |
|---|---|
| T1 design-trace summary | multica-test-t1-design design-trace table |
| T1 case-set link | URL of T1 case set already imported to test-management platform |
| Leader T1 review conclusion | review Pass record (not import authorization) |
Missing any required item → output orchestration "Interrupted: awaiting human to add to Issue comments" table, do not write a coverage conclusion.
Flow
- Read all comments, assemble the 5 items; if missing → interruption list, stop.
- Must read T1 artifact: T1 Confluence
t1-cases.mdlink (comments or JIRA description); T2 only writes delta relative to T1. - For each AC-: implementation evidence (diff/contract)? Can T1 actually verify it? Need new interface cases?
- Signal mapping: new path/field → supplement interface case; new page/popup → supplement functional case; pure refactor → regression risk; AC present but diff absent → implementation gap.
- Confluence landing + automation into repo (see §§ below) → return T2 link.
- Output the template below. Conclusion only:
evaluation completeorevaluation BLOCKED.
Supplemented cases must be written back to T1 Confluence: revise t1-cases → review → republish. After 👤 passphrase, optional XMind/JIRA phase C (see t1-design generation-workflow).
Automation into repo (T2 · Git)
See docs/test-automation-in-repo.md. Read first the MULTICA.md §2 of each repo in the Issue matrix.
| Type | Path source |
|---|---|
| UI Playwright | MULTICA.md UI E2E (usually in frontend repo) |
| API manifest | MULTICA.md API manifest (usually in backend repo) |
Merge into: deploy branch; Leader comment posts the path in MULTICA + commit SHA.
Confluence Landing (T2)
Hard rule: include T1 Confluence link; supplemented CASE writes body (not just a table); no XMind.
Local draft: docs/test/<ISSUE-KEY>/t2-coverage.md
Baseline: multica-platform-confluence/scripts/templates/test-t2-template.md
python multica-platform-confluence/scripts/publish_design.py \
<ISSUE-KEY> docs/test/<ISSUE-KEY>/t2-coverage.md \
--append-jira --json
JIRA block: h3. T2 Coverage Evaluation (Coverage T2).
Platform Collaboration
| Platform skill | Used for |
|---|---|
multica-platform-confluence |
Read T1 child page; publish t2-coverage.md |
multica-platform-jira |
Issue Hub; append link |
multica-platform-apifox |
T2 supplements scenarios by checklist |
T2 material main source is still Issue comment assembly; T1 body uses Confluence t1-cases link as source of truth.
Output Template
# T2 Coverage Evaluation — {JIRA_KEY}
## Conclusion
**evaluation complete | evaluation BLOCKED** — {one line}
## Materials
- AC source / T1 case set / G2 SHA / change list / contract or N/A
## AC Coverage
| AC- | T1 case | Implementation evidence | Status |
|-----|---------|-------------------------|--------|
| | | | covered / false-covered / gap / not delivered / conditional case |
## Case Supplement List
| ID | Action | Required? | Note |
|----|--------|-----------|------|
## New interface cases needed
- none / yes (path + reason). Write target: multica-platform-apifox
## Regression risk
- …
Why it works
T2 only does static gap evaluation, separated from T3 runtime verification; automation assets land in T2, execute in T3.